NVIDIA ASR NIM Microservice API References

The ASR NIM microservice exposes three APIs for speech recognition: an HTTP REST API for simple offline transcription, a gRPC API based on protocol buffers for batch and streaming transcription, and a WebSocket-based realtime API for low-latency interactive use cases.

HTTP REST API

REST endpoints for offline transcription and audio translation. Simple multipart/form-data requests with no gRPC dependency.

Reference

ASR HTTP REST API Reference

gRPC Protobuf API

Protocol buffer definitions for the RivaSpeechRecognition gRPC service, including batch and streaming speech recognition methods.

Reference

gRPC API Reference

Realtime WebSocket API

WebSocket event-based API reference for real-time audio streaming and speech transcription.

Reference

Realtime API Reference